UKF based vision aided navigation system with low grade IMU

When integrating single vision sensor and low grade IMU for 6-DOP navigation, nonlinearity of observation model makes a problem to estimate position, velocity and attitude. Conventional Kalman Filter could not estimate states correctly because it uses linearized model. Due to these reasons, nonlinear estimation should be used to figure out the nonlinear characteristics. By applying Unscented Kalman Filter, this paper copes with the nonlinearity. The estimation performance is demonstrated by numerical simulation. The RMS error of estimated position is analyzed by comparing Extended Kalman Filter results.